Play has several features or functions that drive how play proceeds and derive from game rules and the interests of the players. The purpose of this discussion is to find a theory that describes play and creates a model that can be used as a map — not to find out where we are or where we are going, though these things might be used to invalidate the theory, but rather to find regions of the map that are not yet explored and might drive new game design.
I propose five functions. There may be others. Some might overlap enough to be combined. In principle, any function can apply pressure to any other function.
The game function is act of playing the rules of the game in order to succeed.
The fiction function is the story or narrative that players invent to make sense of their play. In some kinds of play this is literally a story. In others it does not exist or exists as a very thin veneer over the play of the game itself.
The simulation function is play creating an experience that appears to simulate something. A trivial case would be the effort to simulate armed combat. Another might be to simulate the basic forms of Greek tragedy.
The reconstruction function is play that changes the rules of the game.
Risk tension is the degree to which play entails genuine real-world risk. Gambling games emphasize risk tension. Most games de-emphasize it. Very few games attempt to eliminate it.
In all five functions it is acknowledged that different players will desire different weighting and will experience play as having different weights. Solving this social problem might be partially addressed through examination of this theory, but of greater interest to me is the pressures between functions and how those describe play.
To step far afield of role-playing games as a test case, consider the scene in The Deer Hunter where Deniro's and Walken's characters are forced to play Russian Roulette for the amusement of their captors. Now, Russian Roulette has clear rules (game): you put a single bullet in a revolver, spin the cylinder to randomise it, and put the weapon to your head and fire. Clearly we have a pressure chain: game -> risk tension. Deniro's character, however, pleads with his captors to allow three bullets in the chamber. He does this to actually change his risk, so it happens as a result of pressure from the risk tension: and therefore, game -> risk tension -> reconstruction. He does this by taking the weapon loaded with three bullets, now representing very different odds, and shooting his captors (changing the game utterly).
Typical role-playing game play has many chains, but the usual are:
Game -> simulation where play of the game creations an expected simulation result.
Fiction -> reconstruction and simulation->reconstruction where expectations during play for certain kinds of stories or certain qualities of simulation drive ad hoc changes to the rules.
More recently we see games influenced by GNS theory that deliberately create pressures like:
Game -> fiction where playing the game explicitly creates story elements, something that happens outside of the game construction in more traditional games (where you will even find players declaring that the game is irrelevant to the fiction, which indeed it might be in those games as they create fiction indirectly through game -> simulation -> fiction — something that is obvious when you discuss the fiction with these players and they fall back on what the rules allow, what they simulate, and what is realistic to justify or invalidate fiction choices).
The only chains that are strictly under the control of the game designer are those that start with game. So areas to explore in design include might be:
Game -> reconstruction — rules that allow (or require) the change of rules. More correctly, play that, when engaging the rules, creates rule changes. This already occurs in games like Nomic.
Game -> risk tension — adding genuine tension to role playing games by creating situations with real world stakes (physical or emotional). Some folks are already exploring this territory and I expect that they are Scandinavian.
Longer chains are interesting too: Fiction -> reconstruction -> game -> simulation -> fiction happens with some regularity for me: the story demands a revision of game rules to accommodate or realise desired narrative, and the change to the rules now codified creates play that changes what we are simulating or the way (or quality with which) we simulate it.
As stated before, it is certainly the case that players will have different preferences for different functions. Further, they will probably experience the same play as having different weights on different functions. This doesn't strike me as a problem — rather it is just an acknowledgement that people are different and experience play differently. Obviously play that strongly emphasizes or de-emphasizes specific functions will tend to have easily identified proponents and opponents.
The power of this theory, however, is not to divide those people into camps and to arm them. Rather it is to explore the ramifications of strong preference.
Play that explicitly avoids risk tension, for example, does not solely deny risk. It denies all chains that would go through risk tension. This illuminates the fact that strong and non-negotiable preferences have further reaching ramifications than simply denying a single category of play. Instead it inhibits whole branches of the tree of possible function chains. This isn't de facto bad or wrong play, but it would be naive to pretend that it still encompasses all that one expects it too — who knows where these chains lead?
Specifically there is some memetic pressure to reduce the effect of reconstruction in order to better deploy the author's intention (which can only be expressed through game as the rules are the limit of his impact on play). This strikes me as a gross error that not only prunes off huge pieces of a healthy tree but, worse, fails to recognize that this is actually an expected part of play* at many tables and, moreover (and a nod to JDCorley) is basically inevitable. Rules always get changed at the table, though not always deliberately. As well as deliberate reconstruction there is misunderstanding and inadequate clarity in the text. Any effort to design a role-playing game has to at least acknowledge reconstruction and therefore acknowledge any forseeable chains that use it.
So when a table elects to avoid a function of play, they limit available play in ways that are not obvious because now not only does pressure not lead to that function (or more correctly, pressure leading to that function is deliberately unresolved or redirected), it also doesn't go through it to something else. That's not bad, stupid, evil or wrong but rather it might be more than they expect it to be.
Similarly when a designer elects not to support pressure from game to some other node, he limits the capacity to engage that node (by design) but he also indirectly encourages play paths to that node through other ones not under his control. Frequently that path is through reconstruction, denying the author (correctly) absolute ownership of his rules.
* Moreover I will speculate that a significant fraction of the joy that designers feel while playtesting their game and refining it is in fact reconstruction. That they might then deny that part of their play and deliberately exclude it from their delivered game seems baffling unless they have failed to recognize it as an integral part of their experience. It does explain, however, why a game can feel great at the designer's table and be a dead bore for a player armed only with his book and a willingness to take him at his word about the importance of Rules As Written.
When a function generates pressure towards another function it is not necessarily resolved (it might be resisted or it might not be recognized). This seems likely to lead to frustration if the function generating pressure is highly desirable by the player. A strong pressure from fiction towards simulation (a story that needs to be "realised" well) that goes unresolved can result in claims that the game "isn't realistic" or is "too abstract" (though it is in fact the play that isn't resolving the pressure, which may or may not involve the rules). Often it is redirected: the fiction -> simulation pressure continues to reconstruction and is resolved by enabling the simulation through new rules.
This is a case that we confronted directly in development of Diaspora — early on we (I) had hoped to avoid designing with simulation in mind, but in fact there is a lot of pressure towards it from many angles at our table. In the end it was directed through reconstruction and, to some extend risk tension via some very real social friction. We don't like risk tension much so that created other pressures. Eventually some of our reconstruction play wound up in the rules (some as mechanisms that might get ground out as game and some as advice that might simply influence play and perspective).